In realistic, everyday acoustic environments human listeners - regardless of whether they have normal or impaired hearing - depend on the considerable processing power of the human brain to evaluate the auditory scene. The various sources of sound must be parsed, evaluated for relevance and significance, and then attention must be directed to the desired source (the target) while interfering sources (the maskers) are discarded and ignored. Listeners with sensorineural hearing loss (SNHL) - even when wearing hearing aids - often experience extreme difficulty perceptually navigating the auditory scene, severely limiting their ability to communicate effectively. From an acoustic perspective, the designation of a sound source as target versus masker is arbitrary because it depends on the current - and changeable - internal state of the observer. Although the amplification of sounds by hearing aids provides the best (often the only) option for improving communication for listeners with SNHL, current hearing aids inherently fail to solve the source selection problem because they amplify target and masker sounds indiscriminately. The challenge is to devise a hearing aid that focuses only on those sounds the listener chooses to attend and suppresses competing sounds, responding to the wishes of the listener immediately, accurately, and effectively. The goal of the work proposed here is to evaluate a new approach to providing amplification for listeners with hearing loss that is based on the premise that only the listener can make the distinction between which sources to attend and which to ignore. The experiments employ a prototype hearing aid that combines an eye-tracking device with an array of microphones that forms a steerable acoustic beam. By sensing where the eyes are focused, the prototype device can steer the beam of amplification toward the desired source. In that sense, it implements top-down control of focused amplification for the purpose of enhancing sound source selection. The primary goal is to determine the conditions under which top-down control of selective amplification, as implemented by this visually-guided hearing aid (VGHA), can benefit persons with SNHL in complex, dynamic and uncertain listening environments. This goal is to be accomplished under two specific aims that explore 1) hypotheses about top-down control of selective amplification in multitalker sound fields, and 2) the inherent dilemma posed by spatially selective amplification for simultaneously attending to a target source while concurrently monitoring the environment for new sources. A new approach to solving this dilemma will be examined using a dual task. The experimental plan to test these hypotheses employs listeners with hearing loss, matched listeners with normal hearing, and normal-hearing listeners under degraded stimulus conditions simulating the spatial hearing deficits caused by SNHL. Performance under VGHA conditions will be compared and contrasted to that obtained under representative control conditions, and the experimental plan is designed explicitly to take into account the type of masking - energetic versus informational - that is present.